Countercurrent chromatography (CCC), being a support-free liquid-liquid partition chromatographic technique, eliminates the risk of irreversible adsorption of sample components onto the solid support as often observed in conventional liquid chromatography. A variety of existing CCC schemes can be classified into hydrostatic equilibrium and hydrodynamic equilibrium systems. High-speed CCC that is the most advanced hydrodynamic equilibrium system has been widely used for the preparative separation and purification of natural products, while hydrostatic CCC system is efficiently applied to analytical separations using a narrow-bore coiled column arranged around the periphery of the centrifuge bowl in a toroidal form in a seal-free flow-through centrifuge. In this toroidal coil CCC system, however, the retention of the stationary phase is limited to substantially less than 50% of the total column capacity, since the half of each helical turn is entirely occupied with the mobile phase. In order to cope with this problem, a triangular coiled column has been introduced which has improved the retention of the stationary phase to slightly over 40%. Recently various column designs have been introduced to further improve the retention of the stationary phase, including zigzag column, saw tooth column and figure-8 column. In present study, the performance of these three different columns and the traditional coiled column was compared in separation of dipeptides and DNP-amino acid test samples each with a suitable two-phase solvent system using a rotary-seal-free centrifuge fabricated by Pharma-Tech Research Corporation, Baltimore, Maryland, USA. The apparatus holds an aluminum rotary platform measuring about 34 cm in diameter to hold a separation column. The column is made by hooking approximately 17 m long, a 0.46 mm ID FEP (Fluorinated ethylene propylene) (Zeus Industrial Products, Orangeburg, SC, USA) tubing onto the screws upstanding on the rotary platform to form various column configurations such as zigzag, saw toothand 8-figure each with a total capacity of about 2.8 ml. Each terminal of the column is connected to flow tube with a set of tubing connectors. These flow tubes are put together and passed through the center of the central shaft downward and the hollow horizontal shaft of a miter gear, then led upward into the vertical hollow tube support, and finally exit the centrifuge from the center of the upper plate where they are tightly clamped. Two typical two-phase solvent systems including 1-butanol-acetic acid-water (4:1:5, v/v) (BAW) and hexane-ethyl acetate-methanol-0.1 M HCl (1:1:1:1, v/v) (HEMW) were used to separate the dipeptide and DNP-amino acid test samples, respectively. Each solvent mixture was thoroughly equilibrated in a separatory funnel by repeated vigorous shaking and degassing, and the two phases separated shortly before use. Sample solution 1 was prepared by dissolving 25 mg of Trp-Tyr and 100 mg of Val-Tyr in 20 ml of the upper phase of 1-butanol-acetic acid-water, and 40l of this stock solution was used for each separation. Sample solution 2 was prepared by dissolving 5.7 mg of DNP-L-ala, 5.1 mg of DNP-&#946;-ala and 5.3 mg of DNP-DL-glu in 10 ml of the upper phase of hexane-ethyl acetate-methanol-0.1 M HCl (1:1:1:1, v/v), and 40l of this solution was used for each separation. In each separation, the separation column was entirely filled with the stationary phase, either upper or lower phase, followed by sample injection, and the column was rotated at 1000 rpm while the mobile phase was pumped into the coiled column at a given flow rate. The effluent from the outlet of the coiled column was continuously monitored with a Uvicord IIS (LKB, Stockholm, Sweden) at 280 nm and the elution curve was traced using a strip-chart recorder (Pharmacia, Stockholm, Sweden). In order to improve the tracing, ethanol was added to the effluent at the inlet of the detector using a tee connector and a fine mixing tubing (PTFE 0.4 mm ID ca 1 m) at a flow rate of 20 % that of the mobile phase. After the desired peaks were eluted, the run was stopped and the column contents were forced by pressurized air into a graduated cylinder to determine the volume of the stationary phase retained in the column. The stationary phase retention (Sf) was computed by dividing the volume of the retained stationary phase by the column volume and expressed as %. The partition efficiency of separation column was evaluated by computing theoretical plate number (N) for each peak and the peak resolution (Rs) between the peaks using the following conventional equations: N = (4tR/W)2 (1) Rs = 2(t2-t1)/(W1+W2) (2) where tR and W indicate the retention time and the baseline peak width in Eq 1 and those for t of specified peaks in Eq 2. In the present study the performance of the conventional toroidal coil, zigzag, saw tooth and figure-8 columns was investigated in terms of stationary phase retention (Sf), peak resolution (Rs), theoretical plate number (N) and column pressure. The retention of the stationary phase of four different columns in two typical solvent systems, HEMW and BAW was investiged at a flow rate of 0.05 ml/min using the revolution speed of 1000 rpm. DNP-DL-glu, DNP-&#946;-ala and DNP-L-ala were used as test samples for HEMW, and Val-Tyr and Trp-Tyr, for BAW. In general, retention of the stationary phase of the HEMW system was better than that in the BAW except that Sf in the fig-8 column using the lower mobile phase was slightly lower than that in the BAW. When lower phase was mobile phase in the HEMW, the saw tooth column yielded the best Sf at 43.7 % follower by the fig-8 column with Sf at 43.1 % while Sf in the traditional coiled column was worst at 35.7 %. When the upper phase was mobile phase in the HEMW, the fig-8 column gave the best Sf at 55.6 % followed by the saw tooth, zigzag and coiled columns n this order. The Sf in the BAW system showed a similar tendency in both mobile phase groups, i.e., fig-8, saw tooth, zigzag and coiled column in this order. When the lower phase was used as the mobile phase in the HEMW, fig-8 with Rs at 1.69 (1st peak/ 2nd peak) - 1.79 (2nd peak/3rd peak) was best followed by saw tooth with Rs at 1.54 - 1.55, zigzag at 1.36 - 1.29, and coiled column at 1.25 -1.34 (Fig. 3A and Table 1). But when the upper phase was as the mobile phase in HEMW, Rs of the coiled column at 1.60 - 0.88 was better than that of zigzag at 1.35 - 0.73, saw tooth at 1.27 - 0.85, and fig-8 at 1.21 -0.63). In the BAW system, the results of Rs were also similar. When the lower phase was used as the mobile phase, Rs of fig-8 at 1.31 was best among others, whereas in the upper phase mobile the coiled column yielded the best Rs at 1.12. Among all columns tested, the fig-8 column showed the best Rs values in the lower phase mobile. The theoretical plate number (N) of the fig-8 column was also best when the lower phase was the mobile phase indicating that the fig-8 column is highly efficient. The column pressure of all three new columns including zigzag, saw-tooth and fig-8 is much lower than that of the conventional coiled column indicating that a much longer column can be used to further improve the separation The overall results of our experiments indicate that among all columns tested the fig-8 column yielded the best performance when the lower phase was used as the mobile phase. Since the column pressure of all new columns is much lower than that in the conventional coiled column, the partition efficiency of these columns can be further improved by using a longer separation columns.